Open a project folder. Trust it only if you know its contents and origin.
Read the verified extension listing for supported VS Code versions, operating systems and model providers.
For local inference, prepare the model and runtime required by that extension. Memory and disk requirements depend on the selected model.
The OpenMindAI desktop application and a VS Code extension are separate installations. Installing the desktop app alone does not add an extension to VS Code.
2. Install the extension
Open Extensions with Ctrl + Shift + X on Windows/Linux or ⌘ + Shift + X on macOS.
Search for OpenMindAI. Verify the publisher and source repository against the official project announcement.
Open the listing, read its requirements, select Install and reload VS Code if prompted.
If an official release supplies a VSIX instead, open the Extensions menu (…) → Install from VSIX… and select that file.
Use official download sources. To update, open Extensions and choose Update when available. A desktop release version does not necessarily match an extension version.
3. Connect a model
Open the installed extension’s settings from its gear menu → Extension Settings.
Choose a provider supported by its README. For a local provider, start the required runtime and load a compatible model first.
If the extension asks for a server URL, copy the actual endpoint from your runtime. Match its required API format and model identifier; do not assume a port or URL.
If using a hosted provider, enter credentials through the extension’s documented secure setup. Hosted requests send selected context to that provider and may incur charges.
Send a short test message before using a large project. Confirm that the selected model returns a response.
Local inference can work offline once the required files are downloaded if the extension supports it. Installation, downloads, updates and hosted providers require network access.
4. Start a focused coding workflow
Open the extension’s view or use the Command Palette (Ctrl + Shift + P / ⌘ + Shift + P) and search for its registered commands.
Ask about code
Select a small function and ask what it does, which inputs it expects and where it might fail.
Explain a problem
Include the error message, relevant code and expected behavior. Ask for an explanation before applying changes.
Fix and review
Ask for the smallest useful fix, review its diff and run the project’s relevant tests.
Agent workflows
If the installed version supports an agent, review its file and terminal permissions. Check each proposed change and command.
These are suggested workflows. Availability and command labels depend on the installed extension; use its README as the reference.
5. Troubleshooting
No extension appears in search
Confirm the official publisher and listing URL. If no verified listing is available, wait for an official release or use its officially supplied VSIX.
Model connection fails
Check that the runtime is running, the model is loaded and the endpoint matches the extension’s supported API. Review firewall and authentication settings.
First response is slow or times out
Local models can take time to load. Try a short prompt, reduce context or choose a smaller supported model. Change timeout settings only if the extension documents them.
Empty or incomplete response
Test a plain message, confirm model compatibility, then inspect the extension’s Output channel or logs. Update to the latest official extension release.
Commands or panel are missing
Check that the extension is enabled in this workspace and in the correct local/remote environment. Reload the window and check Workspace Trust requirements.
Still stuck?
Share VS Code/extension versions, OS, provider, model, steps and a sanitized error log through Support & Contact. Remove API keys, tokens and private project content.